Differential Gene Expression
Numerous research facets were used to investigate the connection between human CLDN18 and pan-cancer. Based on the integration of the TCGA, GTEx, and HPA datasets, we investigated the distribution and expression of CLDN18 in various tumor cells and tissues. To examine CLDN18 expression in TCGA tumors, we used the R programming language. In STAD, PAAD, LUAD, ESCA, and LUSC, CLDN18 is substantially expressed, as shown in Fig. 1A. Tumor tissues' levels of CLDN18 expression were different from those of normal tissues in part (Fig. 1B). The expression of CLDN18 was increased in the following cell types: CHOL (p < 0.001), COAD (p < 0.001), HNSC (p < 0.001), KIRC (p < 0.001), LIHC (p < 0.001), PAAD (p < 0.05), and PCPG (p < 0.05). Compared to the comparable control tissues, the expression of CLDN18 is lower in the BRCA (p < 0.001), ESCA (p < 0.05), KICH (p < 0.001), LUAD (p < 0.001), LUSC (p < 0.001), THCA (p < 0.001), THYM (p < 0.05), and UCEC (p < 0.001). For some cancers, it didn't seem to make a difference at all.
Clinical Correlation (tumor stage, gender, age)
The findings show that CLDN18 expression varied with tumor stage in ESCA, LIHC, LUSC, PAAD, and TGCT, and was higher in advanced tumor stages compared to early stages in HNSC, LIHC, and THCA. They also show that CLDN18 expression varied with patient age in BRCA, ESCA, LGG, and PAAD, with higher CLDN18 expression in older patients (Fig. 2).
Differences in Gene Activity
We can see that the target genes' activity is strongly expressed in READ, COAD, PAAD, STAD, and LUAD by sorting box line plots (Fig. 3A). With the use of difference box line plots, we can see that the target genes' activity was rated in the BRCA, CHOL, COAD, ESCA, HNSC, KICH, KIRC, LUSC, PAAD, PCPG, THCA, and UCEC with varying levels of expression (Fig. 3B).
Prognostic Value of CLDN18 in Pan‑Cancer
The study found that elevated CLDN18 expression was significantly associated with poorer overall survival (OS) in ACC (p = 0.0437, HR = 2.20477), LGG (p < 0.0001, HR = 3.14771), LIHC (p = 2e-04, HR = 1.94018), MESO (p = 0.0035, HR = 2.03128), and UVM (p = 0.0445, HR = 2.61788), as shown in Fig. 4A. Similarly, elevated CLDN18 expression was significantly related to a poorer disease-specific survival (DSS) in LGG (p < 0.0001, HR = 3.02887), LIHC (p = 0.003, HR = 1.98719), MESO (p = 0.0021, HR = 2.74956), SARC (p = 0.0443, HR = 1.57007), and UVM (p = 0.0303, HR = 3.04985), as shown in Fig. 4B. As shown in Fig. 4C, elevated CLDN18 expression was significantly related to a poorer DFS in COAD (p = 0.0498 HR = 0.41065)、LUAD (p = 0.0251 HR = 0.61149)、STAD (p = 0.0310 HR = 2.13914). In accordance with Fig. 4D, there is a significant correlation between elevated CLDN18 expression and a poorer PFS in ACC (p = 0.0153 HR = 2.19125), LGG (p < 0.0001 HR = 2.09646), PCPG (p = 0.0194 HR = 3.13234), THYM (p = 0.0282 HR = 2.88997), and UVM (p = 0.0249 HR = 2.56116). On the other hand, elevated CLDN18 expression is significantly correlated with a better PFS in SKCM (p = 0.0213 HR = 0.76627) and KIRC (p = 0.0312 HR = 0.70695).
The study analyzed the prognostic value of CLDN18 expression in human cancers and found that higher CLDN18 expression was correlated with poorer Disease Specific Survival (DSS) in LIHC (n = 362, HR = 1.987, p = 0.00296), LGG (n = 504, HR = 3.029, p = 6.24e-08), UVM (n = 80, HR = 3.05, p = 0.0303), SARC (n = 254, HR = 1.57, p = 0.0443), and MESO(n = 66, HR = 2.75, p = 0.00206), as shown in Fig. 5A. In addition, patients with higher CLDN18 expression had poorer OS in MESO (n = 86, HR = 2.031, p = 0.00349), LIHC (n = 370, HR = 1.94, p = 0.000235), LGG (n = 512, HR = 3.148, p = 3.11e-09), ACC (n = 79, HR = 2.205, p = 0.0437), UVM (n = 80, HR = 2.618, p = 0.0445) (Fig. 5B), and also had poor DFS in STAD (n = 215, HR = 2.139, p = 0.031). However, higher CLDN18 expression was related to better DFS in LUAD (n = 302, HR = 0.611, p = 0.0251), COAD (n = 187, HR = 0.411, p = 0.0498) (Fig. 5C). As show in Fig. 5D, higher CLDN18 expression was associated with poorer PFS in UVM (n = 79, HR = 2.56, p = 0.0249), THYM (n = 119, HR = 2.89, p = 0.0282), PCPG (n = 181, HR = 3.132, p = 0.0194), LGG (n = 512, HR = 2.096, p = 7.54e-07), ACC (n = 79, HR = 2.191, p = 0.0153). However, higher CLDN18 expression was related to better PFS in SKCM (n = 457, HR = 0.766, p = 0.0213), KIRC (n = 530, HR = 0.707, p = 0.0312). The above results proved that CLDN18 expression closely related to the prognosis of various cancer types.
CLDN18 Expression and Immune Correlation
Recent research has shown a relationship between immune infiltration and the genesis, progression, and metastasis of human malignancies [16, 17]. To investigate the relationship between CLDN18 expression and various immune cell infiltration in pan-cancer, we used the immuneeconv R software package, which incorporates six cutting-edge algorithms, including TIMER, xCell, MCP-counter, CIBERSORT, EPIC, and quanTIseq (Fig. 6). According to XCELL analysis, the majority of the immune cell infiltration was positively connected with ACC, COAD, ESCA, KICH, LUAD, LUSC, MESO, PRAD, READ, and STAD. In contrast, the majority of immune cell infiltration was inversely linked with KIRP, PAAD, SARC, SKCM, and TGCT. On the other hand, there was no discernible link between BLCA, CHOL, DLBC, THYM, UCEC, and UCS and the majority of the immune cell infiltration values (Fig. 6A). According to CIBERSORT research, CLDN18 was adversely linked with the amount of T cells with CD4 + memory that were resting in most malignancies (Fig. 6B). According to the QUANTISEQ data, CLDN18 was adversely linked with the amount of an unidentified cell infiltration in the majority of malignancies. The correlation between it and T cell regulatory T cells (Tregs), however, was positive (Fig. 6C). Analysis using MCPCOUNTER, EPIC, and TIMER suggested that CLDN18, with the exception of uncharacterized cells, would have a significant positive relationship with the infiltration levels of immunosuppressive cells (Fig. 6D, E and F). These results imply that CLDN18 could function as a novel immunological biomarker for the emergence of tumors.
Correlation of CLDN18 Expression with Immune Checkpoint-related Genes
R was used to evaluate the relationship between immune checkpoint gene expression levels and CLDN18 protein expression levels. Separate correlations between the expression levels of CLDN18 and the 47 immune checkpoint genes were computed. In 33 cancers, there was a positive association between CLDN18 expression and the majority of immunological checkpoints, with significant correlations in ACC, KICH, KIRC, LUAD, LUSC, PRAD, and UVM (p < 0.05). In contrast, the connection between CLDN18 and immunological checkpoints was not significant in the following studies: BLCA, CESC, CHOL, DLBC, THCA, THYM, UCEC, and UCS (Fig. 7).
Correlation Analysis of TMB and MSI
Radar plot revealed that CLDN18 expression was negatively correlated with tumor mutational load in LIHC, LUAD, LUSC, STAD, THCA, UCEC, and UVM and positively correlated with TMB in ACC, BLCA, COAD, ESCA, LAML, LGG, PAAD, SARC, SKCM, TGCT, and THYM (p < 0.05); no significant correlation was found in any other tumors (Fig. 8A). In DLBC, KIRC, LUSC, and STAD, expression of CLDN18 was negatively connected with MSI, whereas in COAD, ESCA, LGG, MESO, READ, SARC, SKCM, and TGCT, it was favorably correlated with microsatellite instability (p < 0.05). In contrast, other malignancies showed no association (Fig. 8B).
GSEA Enrichment Analysis
The study of the signaling pathways that were differentially enriched between the high and low CLDN18 expression groups in the KEGG database using R's GSEA enrichment function helped to clarify the biological roles of CLDN18. The findings demonstrated that several of the pathways, namely Olfactory Transduction, Regulation of Autophagy, Rig I Like Receptor Signaling Pathway, Neuroactive Ligand Receptor Interaction, Cytokine Receptor Interaction, and other pathways, were active in the CLDN18 high expression group (Fig. 9).
Genetic Alteration Differences of CLDN18 in Cancers
We examined the mutation status of the CLDN18 gene using the cBioPortal platform and TCGA data to investigate its role in different malignancies. The high CLDN18 amplification in LUSC (> 9%) and high mutation in UCEC (> 3%) were indicated by a pan-cancer investigation. MESO exhibited a frequency of 1.15 percent, which was the greatest "deep deletion" occurrence. STAD had a frequency of 2.73 percent for "Structural Variant," which was the highest occurrence (Fig. 10A). Figure 10B also depicts the kind, locus, and quantity of CLDN18 genetic variation instances. The majority of the genetic variations were CLDN18 missense mutations, with R55Q/*/L changes found in one patient each with GBM, UCEC, BRCA, and HNSC. The primary CLDN18 gene changes in the cancer group are shown in Fig. 10C.
Mutation feature of CLDN18 in different tumors of TCGA. We analyzed the mutation features of CLDN18 for the TCGA tumors using the cBioPortal tool. The alteration frequency with mutation type (A) and mutation site (B) are displayed. We display the mutation site with the highest alteration frequency (R55Q/*/L). (C)The main type of CLDN18 gene alterations in cancer groups.
DNA Methylation
Based on the TCGA and GEO datasets, the DNMIVD tool was used to examine any possible links between DNA methylation and pan-cancer prognosis. Table 1 displays a summary of CLDN18's 21 methylation sites and associated data. We have chosen a few of the findings shown in Fig. 11. Figure 11A, C, E, G and J, respectively, illustrate the relationship between COAD and CLDN18 gene expression and promoter methylation in CHOL, PAAD, LIHC, and HNSC. We found a strong inverse relationship between CLDN18 DNA methylation and gene expression in CHOL, PAAD, LIHC, HNSC, and COAD, with the exception of LIHC. The association between gene methylation and prognosis in CHOL, PAAD, LIHC, HNSC, and COAD was also shown in Fig. 11B, D, F, H, and K. In CHOL and PAAD, CLDN18 expression with high methylation revealed a poorer prognosis than expression with low methylation. In LIHC, HNSC, and COAD, no statistically significant changes were discovered.
Table 1
DNA methylation of CLDN18.
Gene Symbol | CpG | Group | Relation To Island |
---|
CLDN18 | cg00405112 | Body;TSS1500 | N_Shore |
CLDN18 | cg05487105 | Body | S_Shore |
CLDN18 | cg06580220 | Body | S_Shelf |
CLDN18 | cg09236311 | 3'UTR | OpenSea |
CLDN18 | cg10334121 | Body;TSS200 | Island |
CLDN18 | cg10510478 | Body;TSS200 | N_Shore |
CLDN18 | cg10602180 | Body;1stExon | Island |
CLDN18 | cg10784090 | TSS200 | OpenSea |
CLDN18 | cg11197945 | Body;1stExon | Island |
CLDN18 | cg12938320 | Body;TSS200 | N_Shore |
CLDN18 | cg13002957 | Body;TSS200 | N_Shore |
CLDN18 | cg16167624 | Body;TSS200 | N_Shore |
CLDN18 | cg16240260 | Body | N_Shelf |
CLDN18 | cg17298704 | TSS200 | OpenSea |
CLDN18 | cg17574958 | Body | OpenSea |
CLDN18 | cg18246056 | Body;5'UTR;1stExon | Island |
CLDN18 | cg20366986 | TSS200 | OpenSea |
CLDN18 | cg23008404 | Body;TSS200 | Island |
CLDN18 | cg24269840 | Body;TSS1500 | N_Shore |
CLDN18 | cg24894531 | 5'UTR;1stExon | OpenSea |
CLDN18 | cg25792651 | Body;TSS1500 | N_Shore |
Enrichment Analysis of CLDN18-related Genes
We carried out a screening to find the proteins that bind to CLDN18 and genes that are connected with CLDN18 expression in order to understand the molecular mechanism of the CLDN18 gene in carcinogenesis. We next conducted pathway enrichment studies using these data. We were able to locate 10 CLDN18-binding proteins using the STRING program. The interaction network of the relevant proteins is shown in Fig. 12A. We combined all of the tumor expression data from TCGA using the GEPIA2 algorithm to find the top 100 genes that linked with CLDN18 expression. The expression levels of the genes FAM101A, VSIG1, CYSTM1, CTSE, and FER1L6 were shown to exhibit positive correlations with the expression levels of CLDN18 (p < 0.05), as evidenced by the data displayed in Fig. 12B. As shown in Fig. 12C, the heatmap data showed a positive connection between CLDN18 and the five genes mentioned above across a range of cancer types. An intersection analysis was performed to determine the common members of the two groups, and it showed that CLDN23 was the sole common member (Fig. 12D). The KEGG and GO enrichment analyses of the two datasets were then integrated. According to the data in Fig. 12E and F, the effect of CLDN18 on tumor pathogenesis may be influenced by the manufacture of Mucin type O-glycan. The majority of these genes are connected to biological processes including actin filament organization, plasma membrane organization, and O-glycan processing, according to the GO enrichment study. These genes are associated with the actin cytoskeleton, cell-cell junctions, and apical junction complexes in terms of biological components. Hexosyltransferase activity, acetylglucosaminyl transferase activity, and voltage-gated potassium channel activity, which are all engaged in the ventricular cardiac muscle cell action potential repolarization, are additional molecular activities of these genes.
The Expression Pattern of CLDN18 at Single-Cell Levels
A critical technique for examining the possible roles of candidate molecules at the level of the individual cell is single-cell transcriptome sequencing [18]. Nearly all tumor biological characteristics revealed a negative correlation with CLDN18 expression in breast invasive carcinoma (BRCA). The expression of CLDN18 was inversely correlated with Apoptosis, Cell Cycle, DNA Damage, DNA Repair, EMT, Hypoxia, Inflammation, Invasion, Metastasis, Quiescence, and Stemness in Glioblastoma Multiforme (GBM). In opposition. Positive relationships with angiogenesis, proliferation, and differentiation. Additionally, the results demonstrated that in melanoma, CLDN18 expression was adversely linked with DNA repair (MEL). On the other hand, all others had a positive correlation (Fig. 13A). Additionally, using a T-SNE diagram, CLDN18 expression patterns from individual BRCA, GBM, and MEL cells were shown (Fig. 13B).
Prediction of Immunotherapy Response
The success of patients getting immunotherapy was examined using the GSE78220, GSE67501, and IMvigor210 datasets, as shown in Fig. 14. High CLDN18 expression in GSE78220 was connected to worse and statistically different outcomes of receiving immunotherapy. It was also shown that patients with elevated CLDN18 expression had less successful immunotherapy results in GSE67501 and IMvigor210.
In addition, to evaluate TREM2 expression at the protein level, we analyzed the IHC results provided by the HPA database. As shown in Fig. 15, According to the existing data, CLDN18 may have an oncogenic function in the emergence of several tumor types.